Résumé : Background: The relationship between body mass index (BMI) [weight (kg)/height (m 2)] and serious non-AIDS events is not well understood. Methods: We followed D:A:D study participants on antiretroviral therapy from their first BMI measurement to the first occurrence of the endpoint or end of follow-up (N = 41,149 followed for 295,147 person-years). The endpoints were cardiovascular disease (CVD); diabetes; non-AIDS-defining cancers (NADCs) and BMI-NADCs (cancers known to be associated with BMI in general population); and all-cause mortality. Using Poisson regression models, we analyzed BMI as time-updated, lagged by 1 year, and categorized at: 18.5, 23, 25, 27.5, and 30 kg/m 2. Results: Participants were largely male (73%) with the mean age of 40 years (SD 9.7) and baseline median BMI of 23.3 (interquartile range: 21.2-25.7). Overall, BMI showed a statistically significant J-shaped relationship with the risk of all outcomes except diabetes. The relative risk (RR) for the BMI of <18.5 and >30 (95% confidence interval) compared with 23-25, respectively, was as follows: CVD: 1.46 (1.15-1.84) and 1.31 (1.03-1.67); NADCs: 1.78 (1.39-2.28) and 1.17 (0.88-1.54); and "BMI-NADCs": 1.29 (0.66-2.55) and 1.92 (1.10-3.36). For all-cause mortality, there was an interaction by sex (P < 0.001): RR in males: 2.47 (2.12-2.89) and 1.21 (0.97-1.50); and in females: 1.60 (1.30-1.98) and 1.02 (0.74-1.42). RR remained around 1 for intermediate categories of BMI. The risk of diabetes linearly increased with increasing BMI (P < 0.001). Conclusions: Risk of CVD, a range of cancers, and all-cause mortality increased at low BMI (<18.5) and then tended to increase only at BMI > 30 with a relatively low risk at BMI of 23-25 and 25-30. High BMI was also associated with risk of diabetes.